gbrain-query

What problem does it solve? Agents often waste expensive LLM synthesis calls on simple lookups or skip the knowledge base entirely, producing uncited or stale answers. This Skill defines a mandatory escalation ladder so every brain query starts with fast keyword search and only escalates to heavier tools when needed. ## Core Features & Use Cases - Three-layer escalation ladder: Start with mcp_gbrain_search, escalate to hybrid semantic mcp_gbrain_query, then mcp_gbrain_recall for entity timelines, and finally mcp_gbrain_think for multi-hop synthesis. - Freshness and citation rules: Flags pages older than 90 days on time-sensitive topics and requires slug-plus-date citations like [[people/foo]] in every answer. - Entity-specific lookups: Provides lookup patterns for people, companies, projects, and concepts stored in the brain. - Use Case: When asked "What's the status of the Acme deal?", the agent runs a keyword search first, finds deals/acme-q3 with a high score, and answers with a cited source instead of burning a costly synthesis call. ## Quick Start Ask the agent a question about any entity or topic stored in the brain and it will follow the search-to-think escalation ladder with cited sources.